大数据-代码开发（三）

                                     大数据-代码开发（三）

SparkSQL自定义数据源，并将结果数据下沉到HBase中

import java.util
import java.util.Optional
import com.travel.utils.HbaseTools
import org.apache.hadoop.hbase.TableName
import org.apache.hadoop.hbase.client._
import org.apache.hadoop.hbase.util.Bytes
import org.apache.spark.sql.sources.v2.reader._
import org.apache.spark.sql.sources.v2.writer.{DataSourceWriter, DataWriter, DataWriterFactory, WriterCommitMessage}
import org.apache.spark.sql.sources.v2.{DataSourceOptions, DataSourceV2, ReadSupport, WriteSupport}
import org.apache.spark.sql.types.StructType
import org.apache.spark.sql.{DataFrame, Row, SaveMode, SparkSession}

object HBaseSourceAndSink {
  def main(args: Array[String]): Unit = {
    val spark = SparkSession
      .builder()
      .master("local[2]")
      .getOrCreate()

    val df = spark.read
      .format("com.travel.programApp.HBaseSource")
      .option("hbase.table.name", "spark_hbase_sql")
      .option("schema", "`name` STRING,`score` STRING")
      .option("cf.cc","cf:name,cf:score")
      .load()
    df.explain(true)

    df.createOrReplaceTempView("sparkHBaseSQL")

    df.printSchema()

    val frame: DataFrame = spark.sql("select * from sparkHBaseSQL where score > 60")

    frame.write.format("com.travel.programApp.HBaseSource")
      .mode(SaveMode.Overwrite)
      .option("hbase.table.name","spark_hbase_write")
      .save()

  }
}

class HBaseSource extends DataSourceV2 with ReadSupport with WriteSupport{
  override def createReader(options: DataSourceOptions): DataSourceReader = {
    new HBaseDataSourceReader(options.get("hbase.table.name").get(),options.get("schema").get(),options.get("cf.cc").get())
  }

  override def createWriter(jobId: String, schema: StructType, mode: SaveMode, options: DataSourceOptions): Optional[DataSourceWriter] = {
    Optional.of(new HBaseDataSourceWrite)

  }
}



class HBaseDataSourceWrite extends DataSourceWriter{
  override def createWriterFactory(): DataWriterFactory[Row] = {
    new HBaseDataWriterFactory
  }

  override def commit(messages: Array[WriterCommitMessage]): Unit = {

  }

  override def abort(messages: Array[WriterCommitMessage]): Unit = {

  }
}

class HBaseDataWriterFactory extends DataWriterFactory[Row]{
  override def createDataWriter(partitionId: Int, attemptNumber: Int): DataWriter[Row] = {
    new HBaseDataWriter
  }
}

class HBaseDataWriter extends DataWriter[Row]{

  private val conn: Connection = HbaseTools.getHbaseConn

  private val table: Table = conn.getTable(TableName.valueOf("spark_hbase_write"))

  override def write(record: Row): Unit = {
    val name: String = record.getString(0)
    val score: String = record.getString(1)

    val put = new Put("0001".getBytes())
    put.addColumn("cf".getBytes(),"name".getBytes(),name.getBytes())
    put.addColumn("cf".getBytes(),"score".getBytes(),score.getBytes())

    table.put(put)

  }

  override def commit(): WriterCommitMessage = {
    table.close()
    conn.close()
    null

  }

  override def abort(): Unit = {
    null

  }
}



class HBaseDataSourceReader(tableName:String,schema:String,cfcc:String) extends DataSourceReader  {
  //定义HBase的schema
  private val structType: StructType = StructType.fromDDL(schema)
  override def readSchema(): StructType = {
    structType
  }
  //返回DataReaderFactory
  override def createDataReaderFactories(): util.List[DataReaderFactory[Row]] = {
    import collection.JavaConverters._
    Seq(
    new HBaseReaderFactory(tableName,cfcc).asInstanceOf[DataReaderFactory[Row]]
    ).asJava
  }

}


class HBaseReaderFactory(tableName:String,cfcc:String) extends  DataReaderFactory[Row] {
  override def createDataReader(): DataReader[Row] = {
    new HBaseReader(tableName,cfcc)
  }
}

class HBaseReader(tableName:String,cfcc:String) extends DataReader[Row] {

  private var hbaseConnection:Connection = null
  private var  resultScanner:ResultScanner = null

  private var nextResult:Result  = null

  // 获取HBase当中的数
  val data: Iterator[Seq[AnyRef]] = getIterator

  def getIterator: Iterator[Seq[AnyRef]] = {
    import scala.collection.JavaConverters._
    hbaseConnection = HbaseTools.getHbaseConn
    val table: Table = hbaseConnection.getTable(TableName.valueOf(tableName))
    resultScanner = table.getScanner(new Scan())
    val iterator: Iterator[Seq[AnyRef]] = resultScanner.iterator().asScala.map(eachResult => {
      val str: String = Bytes.toString(eachResult.getValue("cf".getBytes(), "name".getBytes()))
      val score: String = Bytes.toString(eachResult.getValue("cf".getBytes(), "score".getBytes()))
      Seq(str, score)
    })
    iterator
  }
  override def next(): Boolean = {
    data.hasNext
  }
  override def get(): Row = {
    val seq: Seq[Any] = data.next()
    Row.fromSeq(seq)
  }
  override def close(): Unit = {
    hbaseConnection.close()
  }

}